如何为R语言ggplot绘制的多系列散点图添加图例

Hey there! The problem with your current code is that you’re setting the colour parameter outside the aes() mapping for most points—ggplot can’t generate a legend when colors are hardcoded directly like that. Let’s fix this with two straightforward approaches:

Approach 1: Use Tidy Data (Recommended) #

ggplot works best with "tidy" data (one observation per row). First, we’ll reshape your wide-format data into long format using tidyr::pivot_longer(), then plot with a single geom_point() call that automatically creates a legend:

library(tidyr)
library(ggplot2)

# Reshape data to long format
tidy_resid <- residFrame %>%
  pivot_longer(
    cols = c(total_diff, desalination_diff, surfacewater_diff, groundwater_diff),
    names_to = "water_source",
    values_to = "difference"
  )

# Create the plot
testPlot <- ggplot(tidy_resid) +
  geom_point(aes(x = STATEFP, y = difference, colour = water_source, shape = water_source)) +
  # Map water source names to your desired colors
  scale_colour_manual(
    values = c(
      "total_diff" = "red",
      "desalination_diff" = "blue",
      "surfacewater_diff" = "green",
      "groundwater_diff" = "yellow"
    ),
    # Optional: Rename legend labels to be more readable
    labels = c(
      "total_diff" = "Total",
      "desalination_diff" = "Desalination",
      "surfacewater_diff" = "Surface Water",
      "groundwater_diff" = "Groundwater"
    )
  ) +
  # Set all points to use shape=1
  scale_shape_manual(values = rep(1, 4)) +
  xlab('STATEFP') +
  ylab('Difference') +
  ggtitle('Difference for all states', subtitle='For each source')

testPlot

This method keeps your code clean, makes it easier to update later, and aligns with ggplot’s core design principles.

Approach 2: Adjust Your Original Wide-Format Code #

If you prefer not to reshape your data, you can fix the legend by moving the colour mapping inside aes() for each geom_point(), then manually define color mappings with scale_colour_manual():

library(ggplot2)

testPlot <- ggplot(residFrame) +
  # Move colour into aes() with a label for the legend
  geom_point(aes(x = STATEFP, y = total_diff, colour = "Total"), shape = 1) +
  geom_point(aes(x = STATEFP, y = desalination_diff, colour = "Desalination"), shape = 1) +
  geom_point(aes(x = STATEFP, y = surfacewater_diff, colour = "Surface Water"), shape = 1) +
  geom_point(aes(x = STATEFP, y = groundwater_diff, colour = "Groundwater"), shape = 1) +
  # Map legend labels to your desired colors
  scale_colour_manual(
    values = c(
      "Total" = "red",
      "Desalination" = "blue",
      "Surface Water" = "green",
      "Groundwater" = "yellow"
    )
  ) +
  xlab('STATEFP') +
  ylab('Difference') +
  ggtitle('Difference for all states', subtitle='For each source')

testPlot

By putting the legend label inside aes(colour = ...), ggplot recognizes these as categories and generates a legend automatically.

内容的提问来源于stack exchange，提问作者Roelalex1996